CN108652609B

CN108652609B - Heart rate acquisition method and system and wearable device

Info

Publication number: CN108652609B
Application number: CN201810474310.5A
Authority: CN
Inventors: 张方方; 陈维亮
Original assignee: Goertek Techology Co Ltd
Current assignee: Goertek Techology Co Ltd
Priority date: 2018-05-17
Filing date: 2018-05-17
Publication date: 2020-10-02
Anticipated expiration: 2038-05-17
Also published as: CN108652609A

Abstract

The invention discloses a heart rate obtaining method, a heart rate obtaining system and wearable equipment. Therefore, in the application, a proper mu value can be selected according to the motion signal corresponding to the motion state, so that the method is suitable for different motion states, the adaptability is good, filtering can be better performed under different motion states, the filtering effect under different motion states is ensured, pure heart rate signals under different states can be obtained, and the problem that any motion state in the prior art has the same mu value, so that the filtering effect is unstable under different motion states is solved.

Description

Heart rate acquisition method and system and wearable device

Technical Field

The invention relates to the technical field of signal processing, in particular to a heart rate acquisition method and system and a wearable device.

Background

When using the wearable device for heart rate detection, the user may perform different motions, such as walking, running, etc., so that the pulse waves collected by the wearable device are mixed with motion interference signals of different degrees. In order to obtain a pure pulse wave, the motion interference of the acquired pulse wave needs to be eliminated.

Now an adaptive filter as shown in fig. 1 is usually used to eliminate motion interference. The adaptive filter takes noise interference as a processor object, and suppresses or attenuates the noise interference so as to improve the quality improvement of the signal at the output end of the adaptive filter on the signal-to-noise ratio. The adaptive filter is a transversal filter with adjustable tap weight coefficients, M is the tap order, W₁(n)、W₂(n)…W_M(n) represents the value of the tap weight coefficient at the time of n, and the LMS (Least Mean Square) algorithm adopted in the adaptive filter takes the acquired pulse wave signal as a reference channel signal d (n) and takes the acceleration signal as a main channel signalSignal x (n). The optimal criterion for the adaptive filter sampling is the minimum mean square error, which considers the least mean square of the difference between the output signal of the adaptive filter and the desired signal to be optimal. In the starting stage of the self-adaptive state, the tap weight coefficient of the self-adaptive filter is firstly adjusted to carry out self-adaptive training, then the signal on the tap of the filter coefficient is utilized to generate an output signal, and the difference value between the output signal and an expected signal is used for adjusting the weight value through a certain self-adaptive control algorithm so as to ensure that the filter is in the optimal state and achieve the purpose of realizing filtering.

Specifically, W (n) ═ W₁(n)W₂(n)...W_M(n)]^TRepresenting a vector of filter weight coefficients; x (n) ═ X₁(n)X₂(n)...X_M(n)]^TRepresenting an input signal of an adaptive filter; the output signal of the adaptive filter

d (n) represents the expected response or output of the adaptive filter input x (n); the difference between the output signal and the desired signal, i.e., e (n) -d (n) -y (n) -d (n) -W^T(n) X (n). As mentioned above, the LMS algorithm measures the filter quality by using the minimum mean square error, and the formula is as follows: e { E }²(n)}＝E{[d(n)-y(n)]²-iterating the M tap weight coefficients w (n) in order to minimize the mean squared error for optimal performance of the adaptive filter. And setting the tap weight vector of the adaptive filter obtained by the nth iteration as W (n), and setting the mean square error obtained by the nth iteration as e (n), then the weight coefficient obtained by the (n +1) th iteration can be obtained by the following formula: w (n +1) = W (n) + μ e (n) x (n), where μ is an iteration step factor. For the LMS algorithm to converge, the derived value range of μ should be:

wherein λ is_maxIs the maximum eigenvalue of the autocorrelation matrix of the input signal x (n). Using d (n) to subtract the filtered signal y output after the final iteration is completed_end(n) obtaining the pulse wave signal with the movement disturbance removedNumber (n).

As can be known from the principle of adaptive filtering based on the LMS algorithm, the value range of the mu value is the maximum eigenvalue lambda of the autocorrelation matrix of the input signal X (n)_maxIt is decided that in the actual algorithm, if the μ value is calculated in this way for different motion states, the calculation amount of the algorithm is increased by several times or even more, which also means that the power consumption is large and the real-time performance is poor. In the prior art, when the above problems are faced, a fixed μ value is usually adopted for LMS filtering to remove motion artifacts, but the method does not consider that in practical application, motion interference components and a normal heart rate frequency band overlap, and effective heart rate frequency band ranges corresponding to different motion states differ, and if the same μ value is adopted for different motion modes, since the μ value is not suitable for various different motion states, filtering cannot be performed well under each different motion state, filtering effects under different motion states are unstable, and adaptability is poor, so that pure heart rate signals under different motion states cannot be obtained.

Disclosure of Invention

The invention aims to provide a heart rate acquisition method, a heart rate acquisition system and wearable equipment, which are used for selecting a proper mu value according to a motion signal corresponding to a motion state so as to adapt to different motion states, have good adaptability, can perform better filtering in different motion states, ensure the filtering effect in different motion states and further obtain pure heart rate signals in different states.

In order to solve the technical problem, the invention provides a heart rate acquisition method, which comprises the following steps:

establishing a corresponding relation between the motion signal and the iteration step factor mu value;

acquiring a current motion signal and a pulse wave signal;

determining a mu value corresponding to the current motion signal according to the corresponding relation;

and carrying out self-adaptive filtering on the pulse wave signal according to the determined mu value to obtain a heart rate signal.

Preferably, the establishing a correspondence between the motion signal and the μ value includes:

establishing a corresponding relation between the magnitude of the motion signal and the magnitude of the mu value;

establishing a corresponding relation between the motion signal and the motion state;

and establishing a corresponding relation between the motion state and the coefficient of the mu value.

Preferably, the motion signal is an acceleration signal.

Preferably, the establishing a correspondence between the magnitude of each motion signal and the magnitude of the μ value includes:

the acceleration signal has a magnitude of 10⁰At a value of the order of 10^-4；

The acceleration signal has a magnitude of 10¹At a value of the order of 10^-6；

The acceleration signal has a magnitude of 10²At a value of the order of 10^-8；

The acceleration signal has a magnitude of 10³At a value of the order of 10^-10；

The acceleration signal has a magnitude of 10⁴At a value of the order of 10^-12。

Preferably, the motion state comprises at least two of walking, running and riding;

said establishing a correspondence between said motion state and the coefficient of μ value comprises:

when the motion state is walking, the coefficient of the mu value is [1, 3 ];

when the motion state is running, the coefficient of the mu value is [3, 5 ];

and when the motion state is riding, the coefficient of the mu value is [1, 5 ].

Preferably, the process of obtaining the tap order used in the process of adaptively filtering the pulse wave signal according to the determined μ value is as follows:

slave interval

The tap order is selected, wherein,

and the value of a is equal to the value of the acquisition frequency of the acceleration signal, the unit of the acquisition frequency is Hz, and b is not more than the total discrete point number in the sampling time of the acceleration signal.

Preferably, before determining the μ value corresponding to the current motion signal according to the correspondence, the method further includes:

carrying out smooth filtering pretreatment and trend removing algorithm pretreatment on the current motion signal in sequence;

and/or the presence of a gas in the gas,

before the adaptively filtering the pulse wave signal according to the determined μ value, the method further includes:

and sequentially carrying out smooth filtering pretreatment and trend removing algorithm pretreatment on the pulse wave signals.

In order to solve the above technical problem, the present invention further provides a heart rate obtaining system, including:

the establishing unit is used for establishing the corresponding relation between the motion signal and the iteration step factor mu value;

the acquisition unit is used for acquiring a current motion signal and a pulse wave signal;

a determining unit, configured to determine a μ value corresponding to the current motion signal according to the correspondence;

and the filtering unit is used for carrying out self-adaptive filtering on the pulse wave signal according to the determined mu value to obtain a heart rate signal.

In order to solve the above technical problem, the present invention also provides a wearable device, including:

a memory for storing a computer program;

a processor for carrying out the steps of heart rate acquisition according to any one of claims 1 to 7 when executing the computer program.

Preferably, the wearable device is a wristwatch or a bracelet.

The invention provides a heart rate obtaining method, which fully considers that mu values are different in different motion states, therefore, compared with the condition that any motion state in the prior art takes the same mu value, the method can pre-establish the corresponding relation between motion signals and the mu values, each motion signal and the motion state have the corresponding relation, in the subsequent use process, the mu value corresponding to the current motion signal can be determined, and then the pulse wave signals are filtered according to the mu values to obtain pure pulse wave signals, namely heart rate signals. Therefore, according to the method and the device, the proper mu value is selected according to the motion signal corresponding to the motion state, so that the method and the device are suitable for different motion states, the adaptability is good, filtering can be performed well under different motion states, the filtering effect under different motion states is guaranteed, and then pure heart rate signals under different states can be obtained.

The invention also provides a heart rate acquisition system and a wearable device, and the heart rate acquisition system and the wearable device have the same beneficial effects as the method.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed in the prior art and the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.

FIG. 1 is a schematic diagram of an adaptive filter;

FIG. 2 is a process flow diagram of a heart rate acquisition method according to the present invention;

3(a) -3 (e) are in turn acceleration signals of the order of magnitude 10⁰、10¹、10²、10³And 10⁴A graph of the distribution of the magnitude of the μ values of the sample points;

fig. 4(a1) and fig. 4(a2) are schematic diagrams of an original pulse wave signal and an original pulse wave spectrum respectively when the user is in a running state;

fig. 4(b1) and fig. 4(b2) are schematic diagrams of an acceleration signal and an acceleration signal frequency spectrum when the user is in a running state, respectively;

fig. 4(c1) and fig. 4(c2) show that the user is running, and μ ═ 4.6 × 10^-8A schematic diagram of the output signal of the adaptive filter and the output signal spectrum;

fig. 4(d1) and 4(d2) show that the user is running, and μ ═ 4.6 × 10^-8A schematic of the time-filtered heart rate signal, the heart rate signal frequency spectrum;

fig. 4(e1) and fig. 4(e2) show that the user is running, and μ ═ 4.6 × 10^-9A schematic diagram of the output signal of the adaptive filter and the output signal spectrum;

fig. 4(f1) and fig. 4(f2) show that the user is running, and μ ═ 4.6 × 10^-9A schematic of the time-filtered heart rate signal, the heart rate signal frequency spectrum;

fig. 4(g1) and fig. 4(g2) show that the user is running, and μ ═ 4.6 × 10^-7A schematic diagram of the output signal of the adaptive filter and the output signal spectrum;

fig. 4(h1) and fig. 4(h2) show that the user is running, and μ ═ 4.6 × 10^-7A schematic of the time-filtered heart rate signal, the heart rate signal frequency spectrum;

FIGS. 5(a), 5(b) and 5(c) are sequentially statistical distribution graphs of the corresponding μ values of the testers in three exercise states of walking, running and riding;

FIG. 5(d) is a summary plot of μ values corresponding to the sample points of FIGS. 5(a), 5(b) and 5 (c);

FIG. 5(e) is a plot of a summary of μ values corresponding to the sample points of FIGS. 5(a) and 5 (b);

fig. 6(a1) and fig. 6(a2) are schematic diagrams of an original pulse wave signal and an original pulse wave spectrum respectively when the user is in a walking state;

fig. 6(b1) and 6(b2) are schematic diagrams of an acceleration signal and an acceleration signal spectrum when the user is in a walking state, respectively;

FIGS. 6(c1) and 6(c2) are schematic diagrams of the output signal and the output signal spectrum of the adaptive filter when the user is in a walking state and the mu value is determined by the method provided in the present application, respectively;

FIGS. 6(d1) and 6(d2) are schematic diagrams of a filtered heart rate signal and a heart rate signal spectrum, respectively, when a user is in a walking state and a μ value is determined using the method provided herein;

fig. 7(a1) and fig. 7(a2) are schematic diagrams of the original pulse wave signal and the original pulse wave spectrum respectively when the user is in another running state;

fig. 7(b1) and 7(b2) are schematic diagrams of an acceleration signal and an acceleration signal frequency spectrum respectively when the user is in another running state;

FIGS. 7(c1) and 7(c2) are schematic diagrams of the output signal and the output signal spectrum, respectively, of the adaptive filter when the user is in another running state and the μ value is determined using the method provided herein;

FIGS. 7(d1), 7(d2) are schematic diagrams of a filtered heart rate signal, respectively a heart rate signal spectrum, when the user is in another running state, at a μ value determined using the method provided herein;

fig. 8(a1) and fig. 8(a2) are schematic diagrams of an original pulse wave signal and an original pulse wave spectrum when the user is in a riding state, respectively;

fig. 8(b1) and 8(b2) are schematic diagrams of an acceleration signal and an acceleration signal frequency spectrum when the user is in a riding state, respectively;

FIGS. 8(c1), 8(c2) are schematic diagrams of the output signal, respectively, of the adaptive filter when the user is in a riding state, with a μ value determined using the methods provided herein;

FIG. 8(d1), FIG. 8(d2) are schematic diagrams of a filtered heart rate signal, respectively a heart rate signal spectrum, when a user is in a cycling state, at a μ value determined using the methods provided herein;

fig. 9 is a schematic structural diagram of a heart rate obtaining system provided by the present invention.

Detailed Description

The core of the invention is to provide a heart rate acquisition method, a heart rate acquisition system and wearable equipment, wherein an appropriate mu value is selected according to a motion signal corresponding to a motion state, so that the heart rate acquisition method is suitable for different motion states, has good adaptability, can perform better filtering in different motion states, ensures filtering effects in different motion states, and can further obtain pure heart rate signals in different states.

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Referring to fig. 2, fig. 2 is a flowchart illustrating a process of a heart rate obtaining method according to the present invention, the method includes:

s11: establishing a corresponding relation between the motion signal and the iteration step factor mu value;

it should be noted that, the states of a general user include a stationary state and a moving state, and since there is no need to consider the interference of movement in the stationary state, there is no need to perform motion artifact removal on the pulse wave signal, and the detected pulse wave signal can be directly used as the heart rate signal. The heart rate detection method and the heart rate detection device mainly aim at heart rate detection of a user in a motion state. In addition, it should be noted that, in the present application, the adaptive filter based on the LMS algorithm is also used to filter the pulse wave signal, and a very important factor for determining the quality of the LMS algorithm is the adaptive convergence rate, and the speed of the adaptive convergence rate is closely related to the selection of parameters in the LMS algorithm, especially the step iteration factor μ value.

Through extensive research, the application finds that under the condition of a certain iteration step, step iteration factor mu values corresponding to different motion signals such as acceleration signals can be different, and therefore, the appropriate mu is selected for different motion states (corresponding relations between the motion signals and the motion states) in the following processThe method includes the steps of obtaining iteration step-size factors corresponding to different motion signals, specifically, obtaining mu values corresponding to the different motion signals through experiments, obtaining mu values corresponding to the different motion signals through past experience values, and of course, obtaining the mu values corresponding to the different motion signals through other methods. The above mentioned experiment for obtaining the μ values corresponding to different motion signals may be implemented as follows: acquiring different motion signals of a user, and calculating a mu value corresponding to each motion signal, wherein the mu value can be calculated by adopting a calculation method in the prior art, namely according to a maximum eigenvalue lambda of an autocorrelation matrix of the motion signals_maxDetermining the value range of the mu value, wherein,

and then determining a specific mu value from the value range. In order to make the acquired μ values corresponding to different motion signals more accurate, a plurality of users can be selected to perform experiments when the acquired μ values are acquired through experiments, and the final result can refer to a plurality of sets of μ value data.

After obtaining the iteration step factor μ values corresponding to different motion signals, the corresponding relationship between each motion signal and the μ value is established, because the motion signal corresponds to the motion state, the motion state can be determined according to the motion signal, and therefore, the corresponding relationship also exists between the motion state and the μ value. After the corresponding relation between each motion signal and the mu value is obtained, the mu value corresponding to each motion signal can be directly selected through the corresponding relation in the subsequent use process, the calculation amount is small, and each motion state can be well filtered.

In addition, the step is only needed to be executed when the corresponding relation between each motion signal and the mu value is established initially, and the result obtained by the step is directly used without being executed every time unless updating is needed subsequently.

S12: acquiring a current motion signal and a pulse wave signal;

s13: determining a mu value corresponding to the current motion signal according to the corresponding relation;

s14: and carrying out self-adaptive filtering on the pulse wave signal according to the determined mu value to obtain a heart rate signal.

After the corresponding relationship between each motion signal and the mu value is established, before pulse wave is filtered, the current motion signal is firstly obtained, the mu value corresponding to the current motion signal is determined according to the position of the motion signal of the current motion signal in the preset corresponding relationship, and then the acquired pulse wave signal is subjected to adaptive filtering by adopting the determined mu value to obtain a pure pulse wave signal, namely a heart rate signal.

In summary, according to the present application, an appropriate μ value is selected according to a motion signal having a corresponding relationship with a motion state, that is, the μ value is adjusted to adapt to filtering in different motion states, so that on one hand, the computational complexity is reduced, and on the other hand, the adaptability is good, so that filtering can be performed in different motion states well, the filtering effect in different motion states is ensured, and then pure heart rate signals in different states can be obtained.

On the basis of the above-described embodiment:

as a preferred embodiment, establishing correspondence between the motion signal and the μ value includes:

establishing a corresponding relation between the magnitude of each motion signal and the magnitude of the mu value;

establishing a corresponding relation between each motion signal and the motion state;

First, when the μ value is expressed in the form of scientific counting method, the μ value may be expressed as a coefficient order.

Based on this, the application obtains, through research (for example, experiments), that there is a correspondence between the magnitude of the motion signal and the magnitude of the μ value, and there is a correspondence between the motion signal and the coefficient of the μ value, and it is considered that the motion signals in a general range correspond to the same motion state, and the coefficients of the μ values corresponding to the motion signals are similar.

In addition, it should be noted that there are many ways to determine the motion state through the motion signal, for example, when the motion signal is an acceleration signal, the motion state of the acquired pulse wave signal may be classified according to the acceleration signal feature values such as standard deviation, kurtosis, and sample entropy, and the method of determining the motion state may be different for different motion signals, and is determined according to the actual situation. As a preferred embodiment, the motion signal is an acceleration signal.

Specifically, this application discovers through the experiment that acceleration signal can be for the motion state of accurate representation, and acceleration signal and also there is stronger corresponding relation between the mu value, consequently, this application preferentially chooses acceleration signal to establish its and mu value between the corresponding relation, has improved the precision that mu value was selected, and then has guaranteed the filtering effect to the pulse wave signal under the different motion states.

As a preferred embodiment, establishing a correspondence between the magnitude of each motion signal and the magnitude of the μ value includes:

acceleration signal of the order of 10⁰At a value of the order of 10^-4；

Acceleration signal of the order of 10¹At a value of the order of 10^-6；

Acceleration signal of the order of 10²At a value of the order of 10^-8；

Acceleration signal of the order of 10³At a value of the order of 10^-10；

Acceleration signal of the order of 10⁴At a value of the order of 10^-12。

First, in order to acquire the magnitude of the acceleration signal, the acceleration signal needs to be acquired first, and there are many methods for acquiring the acceleration signal in the prior art, and the present application is not particularly limited to how to acquire the acceleration signal. For example, a three-axis acceleration sensor can be adopted, the output 3-axis signals are respectively recorded as an X-axis, a Y-axis and a Z-axis, the algorithm uses the resultant vector as the accelerometer signal, and the formula is as follows:

wherein X, Y, Z in the formula are respectively an acceleration component of an X axis, an acceleration component of a Y axis, and an acceleration component of a Z axis, and the acceleration signals referred to in the present application are all resultant accelerations.

Specifically, the present application obtains the correspondence between the magnitude of each acceleration signal and the magnitude of the μ value through a study (experiment or summary of past empirical values), and specifically refers to fig. 3(a) -3 (e).

Before verifying the accuracy of the embodiment, the LMS filtering effect criterion is briefly introduced:

the LMS filtering effect criterion is evaluated as follows: 1. whether the main frequency of the acceleration signal in the pure pulse wave signal is greatly attenuated relative to the original pulse wave signal or not; 2. whether the main frequency of the pure pulse wave signal is basically consistent with the main frequency of the original pulse wave signal or not.

If the main frequency of the acceleration signal in the pure pulse wave signal is greatly attenuated relative to the original pulse wave signal and the main frequency of the pure pulse wave signal is basically consistent with the main frequency of the original pulse wave signal, the LMS filtering effect is better.

To verify the accuracy of the present embodiment, an example is listed below: the first scenario is: when the user is running, the coefficient of the μ value is determined to be 4.6 (please refer to the next embodiment in principle), and the magnitude of the acquired current acceleration signal is 10²Then, according to the selection rule of the magnitude of μ value in the present embodiment, the magnitude of μ value is 10^-8I.e. determining mu-4.6 x 10^-8。

Referring to fig. 4(a1), 4(a2), 4(b1) and 4(b2), it can be seen that the main frequencies of the pulse wave signal and the acceleration signal are 0.87Hz (marked by a thick line) and 1.758Hz (marked by a thin line), respectively, and due to the interference of the motion (embodied by the acceleration signal) on the pulse wave signal, the original pulse wave signal spectrum has a larger relative amplitude at 1.758Hz as can be seen from the spectrogram of 4(a 2). Using mu-4.6 x 10^-8After the adaptive filter filters the original pulse wave signal, it can be seen from the spectrogram of fig. 4(c2) that the main frequency of the output signal of the adaptive filter is 1.758Hz, which means that the filtered motion interference signal is consistent with the frequency of the acceleration signal. Fig. 4(d2) represents the pure pulse wave signal after removing the motion disturbance, i.e. fig. 4(c2) is subtracted from the original pulse wave signal in fig. 4(a2), and it can be seen from the spectrogram that the amplitude of the pure pulse wave signal at 1.758Hz is reduced by 6.4 compared with the original pulse wave signal, and the main frequency is still 0.87 Hz.

Fig. 4(e2) and fig. 4(g2) give μ ═ 4.6 × 10, respectively^-9And μ ═ 4.6 × 10^-7As a result of the output of the adaptive filter, it can be seen from the spectrogram that the amplitude of the motion noise signal (main frequency 1.758Hz) filtered in fig. 4(e2) is lower than that of fig. 4(c2), which results in less reduction of the frequency of the final pure pulse wave signal (fig. 4(f2)), and the amplitude of the motion noise signal (main frequency 1.758Hz) filtered in fig. 4(g2) is much higher than that of fig. 4(c2), but the main frequency of the final pure pulse wave signal (fig. 4(h2)) changes. This indicates that, under the same conditions, μ ═ 4.6 × 10^-8The best filtering effect is obtained.

In summary, it is easy to see that the method for selecting the magnitude of the μ value provided by the present application can make the finally selected μ value adapt to the motion state at this time, and can obtain a better filtering effect.

As a preferred embodiment, the exercise state includes at least two of walking, running and riding;

establishing a corresponding relationship between the motion state and the coefficient of the μ value, including:

when the motion state is walking, the coefficient of the mu value is [1, 3 ];

when the exercise state is running, the coefficient of the mu value is [3, 5 ];

when the motion state is riding, the coefficient of the mu value is [1, 5 ].

Specifically, the present application divides the exercise state into at least two of walking, running, and riding, and obtains the correspondence between the above exercise state and the coefficient of μ value through research (experiment or statistics on past empirical values). In addition, when a certain motion state is determined, the effect of selecting any one of the μ -valued coefficients in the range of the μ -valued coefficients corresponding to the motion state is almost the same, and can be considered to be the same. For example, when it is determined that the exercise state is walking, the coefficient of μ is [1, 3], and the effect of filtering is almost the same when the coefficient of μ is 1 as compared with when the coefficient of μ is 2 or when the coefficient of μ is 3. Therefore, in practical applications, it may be determined which way to determine the coefficient taking the μ value is used according to actual situations, for example, a lower limit value or an upper limit value or a middle value of a range of the coefficient taking the μ value may be determined, where the middle value may be (upper limit value + lower limit value)/2, and the present application does not particularly limit how to determine the coefficient taking the μ value.

The correspondence between the motion state and the coefficient of the μ value may be specifically obtained as follows:

fig. 5(a), 5(b) and 5(c) show statistical distribution diagrams of corresponding μ values of 7 testers in three exercise states of walking, running and riding, so that the 7 testers respectively test for 5 times in the exercise states of walking, running and riding, the acquired data is subjected to adaptive filtering based on the LMS algorithm to remove exercise trail, and the optimal step factor μ value is counted on the premise of ensuring the filtering effect, namely, the position of the acceleration signal dominant frequency in the pure pulse wave signal frequency spectrum has larger attenuation relative to the original pulse wave signal. FIG. 5(d) is a summary of the μ values corresponding to the sample points of FIGS. 5(a), 5(b) and 5(c) by setting the threshold 1e^-6The step iteration factor μ values can be distinguished from riding and running and walking. FIG. 5(e) is a summary of μ values corresponding to sample points (a) and (b).

To verify the accuracy of the present embodiment, an example is listed below:

the same wearer used corresponding filtering results to the μ values determined with the present embodiment under the exercise of walking (fig. 6(a1), fig. 6(a2), fig. 6(b1), fig. 6(b2), fig. 6(c1), fig. 6(c2), fig. 6(d1), fig. 6(d2)), running (fig. 7(a1), fig. 7(a2), fig. 7(b1), fig. 7(b2), fig. 7(c1), fig. 7(c2), fig. 7(d1), fig. 7(d2)), and riding (fig. 8(a1), fig. 8(a2), fig. 8(b1), fig. 8(b2), fig. 8(c1), fig. 8(c2), fig. 8(d1), fig. 8(d 2)). Taking walking as an example, from the spectrograms (a2) and (b2) of fig. 6, the primary frequency of the original pulse wave signal is 0.78Hz (marked by a thick line), and the primary frequency of the acceleration is 1.58Hz (marked by a thin line), and the amplitude is larger, the output signal obtained by the adaptive filtering process based on the LMS algorithm is fig. 6(c2), the primary frequency of the output signal is 1.58Hz, the finally obtained pure pulse wave signal fig. 6(d2) has larger relative attenuation at 1.58Hz, and the primary frequency of the original pulse wave signal is more obvious. Therefore, the LMS filtering method provided by the application has a good effect. Running and riding have similar effects to walking as described above.

In summary, it is easy to see that the method for selecting the coefficient of the μ value provided by the present application can make the finally selected μ value adapt to the motion state at this time, and can obtain a better filtering effect.

As a preferred embodiment, the acquisition process of the tap order used in the adaptive filtering process of the pulse wave signal according to the determined μ value is as follows:

slave interval

The tap order is selected, wherein,

the numerical value of a is equal to the numerical value of the acquisition frequency of the acceleration signal, the unit of the acquisition frequency is Hz, and b is not more than the total discrete point number in the sampling time of the acceleration signal.

Specifically, the adaptive filter includes tap orders, and the more the tap orders are, the more information of the acceleration signal can be kept, and the better description can be made on the acceleration signal, but if the tap orders are extracted, the better description can be made on the acceleration signalToo many tap orders can cause the increase of calculation amount, so that the selection of the proper tap order is also important, and experiments show that when the tap order is not less than the tap order

And not more than the total number of discrete points within the sampling time of the acceleration signal. It can be seen that the tap order is selected in this way, which can reduce the amount of calculation on the basis of ensuring the filtering accuracy.

In addition, it should be noted that, in order to acquire more acceleration signals so as to ensure a subsequent filtering effect, the acquisition of the acceleration signals usually lasts for a period of time, for example, several seconds, and then the acceleration signals acquired within the period of time are subjected to subsequent processing, which is also the aforementioned acquisition time, and the length of the acquisition time may be determined according to actual conditions.

Specifically, the acquisition frequency of the acceleration signal and the pulse wave signal may be 50Hz, and the time taken for calculating the heart rate is generally between 6s and 10s according to the research result, and specifically, 6s (the above-mentioned acquisition time) may be used for calculating the heart rate. Since the amount of exercise of running is greater than the amount of exercise of walking, running is more than walking in 6s of signal fluctuation times, waveform frequency is high, amplitude range is larger, especially acceleration signal. Since the processing was performed once using data of 6s (300 points) each time, considering that the pulse wave signal of the wearer in the running case is the most complicated, the pulse wave signal acquired in the running case was experimented, and found by calculating λ_maxAnd a proper mu value is selected, so that a good filtering effect can be obtained when the iteration is performed for 200 times. From λ_maxAs can be seen from the calculation principle of (a), in the present algorithm, the μ value is determined by the acquired acceleration signal (x (n)). Meanwhile, under the static condition, the main frequency of the acceleration signal is about 2Hz, the corresponding sampling frequency is 50Hz, and the selected iteration number not only needs to consider the sufficient acceleration signal information and dynamic characteristics, but also needs to consider the calculation amount saving. Considering all the above cases, adaptive filtering based on the LMS algorithm can be usedThe tap order M and the number of iterations in the wave calculation process take values of 20 and 200, respectively. Of course, the tap order and the number of iterations are not particularly limited in the present application, and are determined according to the actual accuracy requirement.

As a preferred embodiment, before determining the μ value corresponding to the current motion signal according to the correspondence, the method further includes:

and/or the presence of a gas in the gas,

before adaptively filtering the pulse wave signal according to the determined μ value, the method further includes:

Specifically, in order to further improve the filtering effect, after the current motion signal is acquired, the μ value is not determined immediately according to the current motion signal, but the current motion signal is preprocessed, specifically, the preprocessing includes smoothing filtering preprocessing and de-trend algorithm preprocessing, wherein the smoothing filtering preprocessing can filter interference signals such as noise, and the de-trend algorithm can be used for eliminating the influence of low-frequency components.

Before the pulse wave signals are subjected to self-adaptive filtering, smooth filtering preprocessing and trend removing algorithm preprocessing are also carried out on the pulse wave signals, and the principle is the same as the above.

In practical application, the current motion signal or pulse wave signal may be subjected to smoothing filtering preprocessing, then to trend algorithm preprocessing, and then to smoothing filtering preprocessing again, so as to improve the filtering effect.

Referring to fig. 9, fig. 9 is a schematic structural diagram of a heart rate obtaining system provided by the present invention, the system includes:

the establishing unit 1 is used for establishing a corresponding relation between the motion signal and the iteration step factor mu value;

the acquisition unit 2 is used for acquiring a current motion signal and a pulse wave signal;

a determining unit 3, configured to determine a μ value corresponding to the current motion signal according to the corresponding relationship;

and the filtering unit 4 is used for carrying out self-adaptive filtering on the pulse wave signal according to the determined mu value to obtain a heart rate signal.

For the introduction of the heart rate obtaining system provided by the present invention, please refer to the above method embodiment, which is not described herein again.

The present invention also provides a wearable device, characterized by comprising:

a memory for storing a computer program;

a processor for carrying out the steps of heart rate acquisition as described above when executing a computer program.

As a preferred embodiment, the wearable device is a wristwatch or a bracelet. Of course, the wearable device is not limited to a wristwatch or a bracelet, and the present application is not limited thereto.

For the introduction of the wearable device provided by the present invention, please refer to the above method embodiment, which is not described herein again.

It should be noted that, in the present specification, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A heart rate acquisition method, comprising:

acquiring a current motion signal and a pulse wave signal;

carrying out self-adaptive filtering on the pulse wave signal according to the determined mu value to obtain a heart rate signal;

the establishing of the corresponding relation between the motion signal and the iteration step factor mu value comprises the following steps:

2. The heart rate acquisition method of claim 1, wherein the motion signal is an acceleration signal.

3. The heart rate obtaining method according to claim 2, wherein the establishing a correspondence between the magnitude of each motion signal and the magnitude of the μ value includes:

Of said acceleration signalOf the order of 10⁴At a value of the order of 10^-12。

4. The heart rate acquisition method according to claim 2 or 3, wherein the exercise state includes at least two of walking, running, and riding;

when the motion state is walking, the coefficient of the mu value is [1, 3 ];

when the motion state is running, the coefficient of the mu value is [3, 5 ];

5. The heart rate obtaining method according to claim 2 or 3, wherein the obtaining of the tap order used in the adaptive filtering of the pulse wave signal according to the determined μ value is:

slave interval

The tap order is selected, wherein,

6. The heart rate acquisition method according to any one of claims 1 to 3, wherein before determining the μ value corresponding to the current motion signal according to the correspondence, the method further comprises:

and/or the presence of a gas in the gas,

7. A heart rate acquisition system, comprising:

the filtering unit is used for carrying out self-adaptive filtering on the pulse wave signal according to the determined mu value to obtain a heart rate signal;

8. A wearable device, comprising:

a memory for storing a computer program;

a processor for carrying out the steps of heart rate acquisition according to any one of claims 1 to 6 when executing the computer program.

9. The wearable device of claim 8, wherein the wearable device is a wristwatch or a bracelet.